A. Popyrin vs R. A. Burruchaga — Prediction & Analysis

TennisEdge's AI model finds no betting value in A. Popyrin vs R. A. Burruchaga (Montreal on 2026-08-03) — the market is efficient, with A. Popyrin favoured at 1.50.

Model probabilities

Independent estimates of who wins, each from a different method. The blend is what the bot prices against the market.

We publish how well these score against the closing market, including where they lose: how accurate are tennis predictions?

Elo ratings

Our own ratings, computed from match results since 2018. Players with fewer than 5 rated matches are omitted rather than shown at a provisional starting value. Last updated Aug. 1, 2026.

Recent form

All Aug. 3, 2026 results & picks →

Full analysis

A. Popyrin (A, 26, ATP 60) vs R. A. Burruchaga (B, 24, ATP 56) — ATP Montreal, hard, 1/64-finals

MODELS vs MARKET
  ELO:        A 68.3%  (model_prob_a = ELO)
  ML:         A 62.3%
  Sim (P-b-P): A 24.5% baseline — see SIM CALIBRATION, inputs judged corrupted
  Betfair ML: 1.62 / 2.60 -> devig A 61.6% (vig 0.2%); comm-adj A = 62.9%, B = 39.7%
  Pinnacle:   1.588 / 2.50 -> no-vig A 61.2% (fetched 15:25Z, fresh); drift_a +0.65pp (605 -> 611) = mild sharp money TOWARD Popyrin
  Book quote  1.50 / 2.62 is SHORTER than the exchange on A — the brief's value_edge 4.7% is a book artifact, not executable value.
  Spread of opinion: ELO 68.3 is the lone number above the two sharp prices; ML and both markets cluster 61-63%.

RECENT FORM (most recent 5)
  Popyrin (A): 08-01 Montreal hard FINAL (quali) W v Kokkinakis 1813 (6-4 2-3, retirement pattern);
    07-29 Vancouver hard L v Andrade 1560 (7-5 3-6 6-1); 06-30 Wimbledon grass L v Lehecka 1803;
    06-22 Eastbourne grass L v Choinski 1524 (lost after winning set 1); 06-16 Halle grass L v Collignon 1615 (straights).
    Also in window: FO L v Svajda 1317; Geneva QF L v Ruud; Rome wins over Berrettini 1832 / Mensik 1752 (early May).
  Burruchaga (B): 07-24 Estoril clay QF L v Blockx 1844 (3 sets); 07-22 Estoril W v Borges 1779;
    07-21 Estoril W v Wawrinka 1552; 07-17 Umag clay SF L v Merida 1772 (straights); 07-16 Umag W v Ugo Carabelli 1822.
    Window also has a win over Cobolli 1915 and a Braunschweig loss to Schoenhaus 1630.

STEP 1b — SHAPE / PATH
  Popyrin: FADING. Since the Rome week he has lost to Choinski 1524, Svajda 1317 and Andrade 1560 — three sub-1600
    opponents in nine weeks, one after leading a set. His only win in six weeks is a Montreal qualifying final that
    the scoreline suggests ended in a retirement. Margins have tightened or collapsed against weak brackets = FADING.
  Burruchaga: STABLE, arguably firm — Umag SF and Estoril QF back-to-back, with wins over Cobolli 1915 and
    Ugo Carabelli 1822 (bracket-above wins). All of it on CLAY.
  Path: Popyrin is at THIS venue and won here 3 days ago = match-sharp, conditions-adjusted, on the match surface.
    Burruchaga's last match anywhere was 07-24 (11 days) and every recent match is clay -> SURFACE TRANSITION onto hard
    plus mild rust. That is the one place Popyrin's price is honestly earned.

STEP 2 — FATIGUE (from brief, precomputed)
  A: 2 matches/14d, 6 sets, 203 min, last 08-01 (3 days rest), back_to_back 0. Age 26 = baseline recovery.
  B: 3 matches/14d, 9 sets, 498 min, last 07-24 (11 days off), nothing in 7d. Age 24 = baseline.
  No extreme spot on either side (no 150min+ yesterday, no 3-in-3, no same-day double). Per the 07-27 calibration a raw
  court-time gap is context only; the 11-day gap makes B the fresher-but-rustier side. Fatigue is NOT an input here
  and appears nowhere in the thesis. Roughly equal.

STEP 3 — NARRATIVE
  Let-down: Popyrin won a round labelled "Final" 3 days ago (08-01) -> POSSIBLE by the letter of the rule, but it is a
    Montreal QUALIFYING final at this same venue, i.e. path sharpness, not a title peak. Treated as neutral, not a fade.
  Venue: Montreal 1-0 in window (that quali final). Title recency: none.
  Surface transition: BURRUCHAGA (clay -> hard). Directional support for Popyrin's ML; explicitly NOT support for any
    tight cover of Burruchaga (transitions widen margin variance both ways).

ELO vs RANKING SANITY
  Popyrin ELO-favourite by 6.7pp over a man ranked FOUR PLACES HIGHER (60 vs 56). Popyrin's rating still carries the
  early-May Rome credit (Berrettini, Mensik) while the last two months are a losing streak against sub-1600 players.
  Verdict: ELO is stale-high here and is discounted; ML (62.3%) and the two sharp prices (61.2 / 61.6%) are the
  trustworthy numbers, and they agree with each other.

SIM CALIBRATION (Step 4b)
  Baseline inputs: A spw 0.571 / rpw 0.383 (50 matches, mixed surfaces incl. clay+grass slump); B spw 0.610 / rpw 0.396
  (only 16 matches, essentially ALL clay at Umag/Estoril/Bordeaux level). Baseline A win prob 24.5% — a 37pp divergence
  from ELO/ML/market, driven by an off-surface, weak-field, thin-sample SPW for B and a slump-depressed SPW for A.
  Calibration attempt: spw_a pushed to the clamp limit (baseline 0.571 -> used 0.621, clamped:true) lifts A only to 48.9%.
  The +-0.05 clamp CANNOT reconcile the sim with the consensus -> the simulation is outside its competence for this
  matchup. Per Step 4b step 3 that is a finding: NO margin bet may be staked off this uncalibrated distribution.

MARKET CHOICE
  ML A (Popyrin) @1.62 BF: comm-adj implied 62.9%. Honest probability ~62% (ML 62.3, Pinnacle 61.2, BF devig 61.6,
    ELO discounted as stale-high, sim dissenting hard the other way). Edge ~ -0.9pp. Min odds needed at p=0.62:
    1.694 full-ladder / 1.661 probe. 1.62 is BELOW both -> automatic PASS on pricing discipline alone.
  ML B (Burruchaga) @2.60 BF: comm-adj implied 39.7%; my prob ~38%. Negative. The only number arguing B is the sim,
    whose inputs I have just declared corrupted — I cannot stake an edge from a distribution I rejected.
  Total 22.5: BF under 1.95 (comm-adj 52.6%), sim under 57.7% -> nominal 5.1pp. KILLED twice over: (a) it comes from the
    uncalibrated distribution above, (b) sharp cross-check — Pinnacle 1.98/1.893 devigs to under 51.1%, and BF devigs to
    50.6%; both markets say ~51%, the sim's 57.7% is the lone outlier. Sharp contradiction of ~6.6pp = stale quote, not value.
  Handicap: BF line_a -2.5 @1.95 / B +2.5 @1.98; Pinnacle -2.5 @1.917/1.952 -> sharp cover A 50.5%, B 49.5%; BF ~identical.
    alt_markets.model.handicap is null and the resim distribution is uncalibrated, so there is no defensible cover estimate.
    AH: B +2.5 @ 1.98 (pass — no calibrated cover anchor; not a dog-ML-at-3.0+ spot).
  refutation_check: not run — PASS decision, no web access permitted for passes.

GATES
  G1 edge-origin: FAIL — honest ML edge vs the executable Betfair price is ~-1pp (need +5%, or +3% probe); the only
    number producing a nominal edge is a stale-high ELO that ML, Pinnacle, Betfair and live form all contradict.
  G2 consensus:   [tier: capped] — would be lone-signal at best (ELO alone, contaminated by May results); moot at G1 FAIL.
  G3 fatal-risk:  [none] — no unencodable risk; the margin markets are blocked by the numeric bar and the sharp
    cross-check, not by a story veto.

VERDICT — PASS
  Two sharp prices, the ML model and (violently) the simulation all sit at or below the executable exchange price for
  Popyrin, and 1.62 is short of the 1.694 minimum my own probability demands. The one genuine edge in the match is
  qualitative and already priced: Burruchaga is stepping off pure clay onto hard after 11 days idle while Popyrin is
  match-sharp at the venue. Against that, Popyrin has lost to a 1524, a 1317 and a 1560 in the last nine weeks — a
  FADING favourite at a short price. Those two cancel to roughly the market number. The under 22.5 was the only market
  showing a nominal 5% edge and it evaporates on the sharp cross-check plus the failed sim calibration. Nothing to bet.